A distributed content analysis network uses the processing capabilities of customer-premises equipment as subordinate nodes for analyzing multimedia programs. A master node selects a program and identifies subordinate nodes that are available for analysis, which may include both nodes tuned to the program and idle nodes. The master node divides the program into segments for analysis and instructs each subordinate node to analyze a segment. The subordinate nodes then provide analysis results back to the master node, which may build a metadata profile for the program based on the analysis.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A master node in a distributed content analysis network, the master node comprising: a memory device programmed with instructions, executable by a processor wherein the instructions, when executed by the processor perform operations, the operations comprising: identifying a program in a multimedia content stream provided to a plurality of subordinate nodes; dividing the program into discrete segments, wherein each of the plurality of subordinate nodes is associated with a corresponding discrete segment that is different; assigning analysis tasks to the plurality of subordinate nodes, wherein each of the analysis tasks instructs a different one of the plurality of subordinate nodes to respectively analyze the corresponding discrete segment of the program; receiving results of the analysis tasks from the plurality of subordinate nodes; and generating a metadata profile of the program based, at least in part, on the results.
A master server in a distributed network analyzes multimedia content. It identifies a program within a stream sent to multiple subordinate devices (e.g., set-top boxes). The server splits the program into distinct segments and assigns each segment to a different subordinate device for analysis. Each device analyzes its assigned segment and sends the results back to the master server. The master server then creates a metadata profile for the program based on these combined results. This allows for distributed processing to analyze multimedia content.
2. The master node of claim 1 wherein the corresponding discrete segment of the program is a spatial segment of a video component of the program as displayed on a screen.
The master server described in the previous claim analyzes multimedia content by dividing a program into segments. In this variation, the segments are spatial portions of the video displayed on the screen (e.g., dividing the screen into regions and analyzing each region separately). The analysis tasks are performed on these different spatial areas, providing a more granular video analysis.
3. The master node of claim 1 wherein the corresponding discrete segment of the program is a temporal segment of the program.
The master server described earlier analyzes multimedia content by splitting a program into segments. In this case, the segments are temporal portions of the program (i.e., dividing the program into different time intervals). Each subordinate node analyzes a specific time frame, allowing for analysis of how content changes over time.
4. The master node of claim 1 wherein the analysis tasks perform pattern matching on the multimedia content stream.
The master server described earlier analyzes multimedia content by splitting a program into segments and assigning analysis tasks to subordinate nodes. In this version, the analysis tasks involve pattern matching on the multimedia content stream. This means the subordinate nodes search for specific patterns or features within their assigned segment.
5. The master node of claim 4 wherein the pattern matching includes performing facial recognition to determine a match among a set of predetermined faces.
The pattern matching analysis described earlier, where a master server splits a program into segments and analyzes them, specifically performs facial recognition. The subordinate nodes attempt to match faces within their assigned segments against a predefined set of known faces. This enables identification of individuals appearing in the multimedia content.
6. The master node of claim 4 wherein the pattern matching includes parsing an audio component of the program and performing voice recognition on the audio component to determine a match among a set of predetermined voices.
The pattern matching analysis described earlier, where a master server splits a program into segments and analyzes them, specifically involves parsing the audio and performing voice recognition. Subordinate nodes analyze the audio to identify specific voices from a predefined set of voices.
7. The master node of claim 4 , wherein the operations further comprise: performing speech-to-text conversion on a portion of an audio component of the program, wherein the pattern matching is performed on text phrases resulting from the speech-to-text conversion.
The pattern matching analysis described earlier, where a master server splits a program into segments and analyzes them, also performs speech-to-text conversion on the audio. The pattern matching is then performed on the resulting text phrases, enabling identification of spoken keywords or phrases within the multimedia content. This allows analysis of spoken content even without pre-defined voice models.
8. The master node of claim 4 wherein the pattern matching includes parsing a closed-captioning component of the program and performing pattern matching on captions resulting from the parsing.
The pattern matching analysis described earlier, where a master server splits a program into segments and analyzes them, includes parsing the closed-captioning component. Pattern matching is then performed on the parsed captions, enabling analysis of textual content accompanying the multimedia.
9. The master node of claim 1 wherein the program is a sport contest and the analysis tasks include instructions to characterize events occurring within the sport contest.
The master server analyzing multimedia content, as described earlier, is applied to sports contests. In this case, the analysis tasks performed by the subordinate nodes are designed to characterize specific events occurring within the sport contest (e.g., identifying goals, fouls, or player actions).
10. The master node of claim 1 wherein at least one of the plurality of subordinate nodes is tuned to the program.
In the system where a master server splits a program into segments and assigns them for analysis, at least one of the subordinate nodes is already tuned to the program being analyzed. This node can immediately begin analyzing its assigned segment without needing to switch channels or streams.
11. The master node of claim 1 wherein the plurality of subordinate nodes includes an idle subordinate node.
In the system where a master server splits a program into segments and assigns them for analysis, at least one of the subordinate nodes is an idle node. This means the node is not currently processing other content and can be dedicated to the analysis task.
12. The master node of claim 1 wherein the plurality of subordinate nodes comprises customer premises equipment.
The system where a master server splits a program into segments and assigns them for analysis uses customer premises equipment (CPE) as the subordinate nodes. This means the analysis is performed on devices located in users' homes or businesses.
13. The master node of claim 12 wherein the customer premises equipment comprises a set-top box.
The system where a master server splits a program into segments and the analysis is performed on customer premises equipment, uses set-top boxes as the CPE. The set-top boxes analyze assigned segments of the multimedia content and send results back to the master server.
14. A method of analyzing a multimedia content stream, the method comprising: transmitting the multimedia content stream to a plurality of subordinate nodes, wherein a master node and the plurality of subordinate nodes communicate in a distributed content analysis network; identifying, by the master node, a program in the multimedia content stream for analysis; dividing, by the master node, the program into discrete portions, wherein each of the plurality of subordinate nodes is assigned a corresponding discrete portion that is different for each subordinate node; distributing analysis tasks to the plurality of subordinate nodes, wherein each of the analysis tasks instructs a different one of the plurality of subordinate nodes to respectively analyze the corresponding discrete portion of the program; receiving analysis results from the plurality of subordinate nodes, wherein the analysis results indicate a property of the program; and generating a metadata profile, based at least in part on the analysis results, for association with the program.
A method for analyzing multimedia content involves distributing the content to several subordinate devices within a network. A master server identifies a program within the content stream and divides it into distinct segments. Each subordinate device is assigned a different segment to analyze. The master server distributes analysis tasks to these devices, instructing them to analyze their assigned segments. The devices send back analysis results, which indicate properties of the program. The master server then generates a metadata profile for the program, based on these results.
15. The method of claim 14 , further comprising: configuring the plurality of subordinate nodes to: receive the multimedia content stream; perform the analysis tasks received from the master node to analyze the corresponding discrete portion of the program; and provide the analysis results to the master node.
This method builds on the previous description where a master server distributes multimedia content and analysis tasks to subordinate nodes. The subordinate nodes are configured to receive the content stream, perform the specific analysis tasks received from the master server on their assigned segment, and then provide the results back to the master server.
16. The method of claim 14 wherein the corresponding discrete portion of the program is a temporal segment of the program.
The method described earlier, where a master server splits a program into segments and assigns them for analysis, uses temporal segments of the program as the distinct portions. Each subordinate node analyzes a different time frame within the program.
17. The method of claim 14 wherein the corresponding discrete portion of the program is a spatial segment of a video component of the program.
The method described earlier, where a master server splits a program into segments and assigns them for analysis, uses spatial segments of a video component as the distinct portions. Each subordinate node analyzes a different area of the video frame.
18. The method of claim 14 wherein the corresponding discrete portion of the program is an audio component of the program.
The method described earlier, where a master server splits a program into segments and assigns them for analysis, uses the audio component of the program as the distinct portion. One or more subordinate nodes are assigned to analyze sections of the audio.
19. The method of claim 14 wherein the corresponding discrete portion of the program is a closed captioning component of the program.
The method described earlier, where a master server splits a program into segments and assigns them for analysis, uses the closed captioning component of the program as the distinct portion. Subordinate nodes analyze the closed captions for particular information or patterns.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
December 23, 2008
July 23, 2013
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.